Representation and the realist-constructivist controversy

Human Studies 11:117-132 (1988) © Kluwer Academic Publishers

Representation and the realist-constructivist controversy

PAUL TIBBETTS

Department of Philosophy, University of Dayton, Dayton, OH 45469, USA

1. Introductory remarks: The realist constructivist debate

There are at least three issues surrounding the problem of representation: (1) The representational device or RD (e.g., maps, electrical diagrams, chemical formulae, models, etc.) and the extent to which such RDs are socially constructed, interpreted and deployed; (2) the ontological status of the represented object (RO); and (3) questions concerning the accuracy with which (1) represents (2). For the realist, RDs ultimately denote some independently existing non-cognitive structure or process. Realists recognize the constructivist dimension to RDs and the extent to which such devices and their use are defined by inquirer-contingent criteria, though they insist that RDs ultimately have to map onto some inquirer-independent, real-world properties. For realists, if this were not the case then RDs would represent nothing at all and therefore the data points provided by RDs would be unintelligible. (For variations on the realist position see: Bhaskar, 1978; Jarvie, 1983, 1984; Laudan, 1977, 1981; and Popper, 1972.)

One traditional problem with such realist accounts concerns just what it is RDs represent (issue (2) above). A pragmatic response to this query is that this is an issue best left to metaphysi- cians and those concerned with questions of ontology. What does matter - on this line of thinking - are the data points, particularly those points that: (i) are consistent with theory, (ii) are theoretically interesting, (iii) occasion theoretical revision or ex- tension, and (iv) have heuristic value for further research. After all, why would the realist - or anyone else for that matter - be

118

remotely interested in the relation between RDs and ROs, and the data points the latter occasion in the former, unless there were some theoretical or pragmatic significance associated with RDs? And, continuing this line of response, neither can relativists and constructivists in the sociology of science ignore the fact that RDs do generate reams of non-random data points.

Of course, constructivists counter that what constitutes theoretically significant as against non-significant, or non-random as against random, data points requires reference to epistemic criteria posited by a given community of inquirers. Such criteria include theoretical consistency, heuristic value, predictive accuracy, and Lakatos' (sophisticated) falsifiability. Apart from such negotiated and therefore contingent evaluative criteria, estimates of significance or non-significance are unresolvable in principle. (Cf. Barnes and Edge, 1982; Bohme, 1977; Collins, 1983a, 1983b, 1985;and Knorr-Cetina and Mulkay, 1983.) Still, realists can counter, such socially-constructed epistemic criteria must be criteria for evaluating something. And so the issue bounces back and forth between realist and constructivist. Understandably, a number of sociologists of science have increasingly turned their attention to this debate between realists' and constructivists' accounts of linguistic and non-linguistic representation in science. However, rather than pursuing this issue on a relatively abstract, philosophical level, these sociologists of science examine representational practices within the day-to-day context of inquiry and the interpretive controversy surrounding such practices. (Cf. Collins and Pinch, 1982; Garfinkel, Lynch and Livingston, 1981; Grenier, 1983; Knorr-Cetina, 1981 ; and Pickering, 1984.)

Nor is one necessarily committed to either a realist or a constructivist account in a mutually exclusive sense; elements of both in fact appear in most writings on the subject. The contrast between constructivism and realism is the emphasis respectively given - or not given - to the social contingencies surrounding RDs and associated evaluative criteria, and the supposed epistemic independence of the data points from such considerations.

In the remainder of this paper I examine some recent accounts in the sociology and philosophy of science literature regarding RDs, the R D - R O relation and the issue of data points, show- ing the extent to which realist and constructivist elements are

119

co-present in these accounts. To focus our discussion, Section 2 below examines the account of RDs and the R D - R O relation in the writings of Barry Barnes. As will be seen, Barnes' position in- corporates both realist and constructivist themes. Particularly problematic in Barnes' account is his use of a pictorial model of representation regarding the R D - R O relation. The issue is whether such a model is consistent with a constructivist programme.

My thesis throughout is that to the extent the realist-constructivist debate regarding representation in science is posed as an e i the r -o r situation between mutually exclusive and contrary op- tions, the debate is a red herring. A few recent writers appear to recognize this; too many do not. For this writer, the salient issue is the extent to which realist and constructivist elements are mutually at work and interactive in the design and utilization of RDs in scientific contexts.

2. One constructivist account of representation

In Interests and the Growth o f Knowledge (1977), Barnes examines various instances of RDs, with emphasis on the socially- constructed and socially-defined relation between these RDs and a given represented object or RO. Barnes' examples of RDs include: diagrams, a City map, a medical illustration of the musculature of the human arm, and contour lines on an elevation map. The issue Barnes addresses concerns the socially constructed and defined representational accuracy between RD and RO and the extent to which criteria of accuracy are socially contingent. In addition, for Barnes, considerations of accuracy are not independent of users' "procedures, competencies, techniques and objectives." As Barnes (1977:6)remarks ,

Representations are actively manufactured renderings of their referents, produced from available cultural resources. The particular forms of construction adopted reflect the predictive or other technical cognitive functions the representation is required to perform when procedures are carried out, competencies executed, or techniques applied. Why such functions are initially required of the representation is generally intelligible,

120

directly or indirectly, in terms of the objectives of some social group.

I fully concur with Barnes' account of symbolic representations (wiring diagrams, maps, etc.) as 'analogous to techniques' rather than as merely objects of passive contemplation. Any account which divorces RDs from the contexts of praxis that define and concretely situate such devices clearly ignores a salient - p e r h a p s the salient - influence on the construction and utility of RDs. (For further discussion of this point I highly recommend Hacking's Representing and Intervening, 1983; esp. 130-146. )

Unfortunately, the examples of ROs Barnes specifically employs are physically static and spatially fixed: circuit boards, built environments, musculature and elevations are non-dynamic structures. Barnes apparently takes such structures and their RDs as paradigmatic of the representational relation. For example, terrain maps - surely a case of a static RD of a static RO - constitute for Barnes (1977:7), "one of the clearest and most accessible contexts in which to examine the connection between the structure of representations and their funct ion." Consequently, the static examples of RDs Barnes cites by no means exhaust t h e range of such devices in everyday or scientific inquiry. As suggest- ed above, Barnes' examples do not extend to: (i) ROs in a dynamic, non-static state or (ii) temporally-changing events and processes. Given (i) and (ii), the static character of maps, schematic diagrams and textbook illustrations can not in principle adequate- ly represent change - except, of course, through a series of such diagrams or illustrations. For example, the terrain effects of ero- sion, aircraft performance or urban expansion can be expressed by a series of maps at different time frames, but the dynamics and continuity of such changes are not captured. Even more problematic for Barnes' account is when there are no spatially- defined objects to represent (e.g., airfoil lift-drag coefficients, metereological phenomena, unemployment rates or demographic changes).

Perhaps the problem with Barnes' analysis of RDs is that it un- necessarily restricts itself to a picturing model of the R D - R O relation. As Barnes (1977:4) remarks,

121

Our strategy should be to reveal pictorial representations ... as typical of representation and knowledge generally.

Again (1977: 7), as pictorial representations,

Maps indeed afford one of the clearest and most accessible contexts in which to examine the connection between the structure of representations and their function.

In many of the examples Barnes employs, the RD maps onto structural features of the RO. Such mapping relations are the easiest representational relations to deal with, so it is understand- able why Barnes focused on these particular relations. However, the mapping model does not work where spatial and contour features of the RO are either absent or irrelevant. For instance, an EKG record does not in any literal/representational sense provide a topographical-like map of the electrical activity of the heart in the way that Barnes' wiring diagram does of an electrical circuit. Nor does a chemical formula map onto molecular structure in any obvious topological fashion! Given these counter-examples it is far from clear that pictorial representation is "typical of representation and knowledge generally," or that maps constitute a paradigm case of RDs.

Admittedly, some RDs do map onto their referents in a fashion. For example, the RDs (arrows, curving lines, numbers, etc.) used on weather maps do depict highs and lows, cold and warm fronts, wind direction, pressures, etc. But it should not be concluded that this constituted a one-to-one mapping between RDs and real (weather) events. On the contrary, such RDs are not givens but constructs. The history of measuring, encoding and depicting weather phenomena shows that meteorologists' RDs are but shorthand devices for representing barometric, tempera- ture and wind speed/direction measurements. Clearly the con- cept of an isobar (a line connecting equal pressures), rather than literally representing something, was simply borrowed from the fiction of contour lines (equal elevation) on geological maps. What then do isobars (or contours lines) map onto? Nothing. On a constructivist account they are simply useful conceptual devices for introducing order into measurements and, in turn, for extrapo-

122

lating from past to future data points. Consequently, Barnes' account in Interests and the Growth o f

Knowledge deals with a relatively narrow and non-representative range of RDs, namely, those purPortedly mapping onto structural features of spatially and temporally static objects. Furthermore, in contrast with questions o f utility, the fidelity with which RDs picture their corresponding ROs may be the least interesting issue to raise regarding scientific representation.

In any case, and in spite of the serious conceptual problems noted in the preceding paragraph, Barnes' account of representation clearly endorses a constructivist version of the creation and deployment of RDs. This follows given that for Barnes all RDs are "actively constructed assemblages of conventions or meaning- ful cultural resources ..." (1977:9). Again, the significance of RDs - and of scientific knowledge in general - necessarily requires reference for Barnes to 'institutionalized technical procedures' (1977: 9; emphasis added).

3. The realist position regarding represented objects

As to the nature of the RO, Barnes never seriously doubts that there is a representation-independent reality which ultimately grounds RDs. The reason why scientific knowledge - and empirical knowledge in general - is useful "is precisely because the world is as it is" (1977:10; emphasis added). Consequently, knowledge is "a function of what is real, and not the pure prod- uct of thought and imagination." Elsewhere, Barnes (1977:25- 26) adds that there is a single reality which is mind and culture independent and which can not be symbolically captured in any finite set of RDs.

Everything of naturalistic significance would indicate that there is indeed one world, one reality, 'out there', the source of our perceptions if not their total determinant ... Reality is the source of primitive causes, which, having been pre-processed by our perceptual apparatus, produce changes in our knowledge and the verbal representations of it which we possess.

123

Bhaskar (1981:363) also claims that the extent to which scientific theories are cumulative and amenable to revision,

strongly indicates such theories are (fallible) at tempts to describe real states and structures of Nature, as they succeed one another in providing better accounts of a mind-independent reality.

In a similar realist vein, Bloor ( 1 9 7 6 : 3 6 - 3 7 ) presupposes the existence of a common external world 'with a determinate structure. ' This external world

is assumed to be the cause of our experience, and the common reference of our discourse ... Often when we use the word ' t ruth ' we mean just this: h o w the wor ld s tands ... In particular the assumption of a material world with which men establish a variety of different adaptations is exactly the picture presup- posed by the pragmatic and instrumental notion of correspondence ...

Obviously, neither Barnes, Bhaskar nor Bloor would subscribe to that (extreme) constructivist account which argues that the so- called 'mind- and culture-independent reality' of the realists (particularly re. micro-level phenomena) is: (i) a mere conceptual posit, an extrapolation from theory-interpreted data points, and therefore (ii) a digression (or even regression) into speculative metaphysics. Either we remain silent about this supposed independent reality (which is Wittgenstein's proposal in the last line of the Tractatus: 'Whereof one cannot speak one must remain silent'), or we proceed to describe in some detail just what this reality consists of. The problem, though, with this latter tack is that even if we employ our 'best-informed and current scientific knowledge' our descriptive accounts of reality will be contingent on the explanatory Kuhnian-like paradigms currently in vogue. Accordingly, what inquirers are prepared to posit as a real enti ty or causal relation today may be demoted to artefactual status tomorrow. So, when is this inventory of reality to be taken? Certainly not today; perhaps only in Peirce's 'long run.' (As Peirce, 1885:8.41, once remarked: "the real is that which any

124

man would believe in, and be ready to act upon, if his investiga- tions were to be pushed sufficiently far.")

This reply does not of course conclusively prove that the realism of Barnes, Bloor or Bhaskar is conceptually untenable. Still, consideration of the interpretive and constructive dimension in scientific inquiry makes one hesitant to endorse any ontology which grounds itself in 'the best available scientific evidence.' Such a realist reconstruction also presumes that scientific knowledge is cumulative, progressive and approximates to reality over time. Obviously, such an assumption begs-the-question concerning an independent reality! Finally, such a realist ontology could also be said to be irrelevant and empty baggage given that it tells us nothing whatsoever regarding the day-to-day interpretive, social and negotiative activities associated with the generation and legitimation o f RDs and ROs.

In their account of knowledge construction and legitimation in science, realists consistently ignore such inquirer-contingent considerations as largely irrelevant to epistemological matters proper. Unfortunately for this line of thinking, to understand the conceptual career of RDs and ROs no sharp demarcation can in fact be drawn between the (sociological) context of discovery and the (epistemological) context of justification. Woolgar (1983:246) provides an account of how realists would submerge the interpretive]constructivist dimension in inquiry:

an initial use of accounting procedures to constitute a new reality is subsequently regarded as no more than an attempt to report upon or reflect what was there all along ... [According- ly,] the reversal of the connection between account and object also entails the removal from the scheme of any constitutive activity of the discoverer.

In Art and Artifact in Laboratory Science, a recent study of scientists' situated laboratory practices and the achievement of order, Lynch (1985:202-216) concluded that: (i) inquirers' technical talk and technical work serve to establish and define objectivity, and apart from such considerations no independent a priori standard of objectivity would be available to inquirers in evaluating 'real-world objects,' (ii) in opposition to a realist

125

account, the object of reference (the RO) in inquirers' situated discourse is not an integral phenomenon with a unified meaning, (iii) rather than 'objectivity ' being lost in the dialectical exchange between inquirers' differing accounts, such disagreement and eventual convergence serves to establish and define objectivity, (iv) the objectivity of the RO is therefore constructed over time rather than an ontological given and, consequently, (v) scientific accounts re. the objectivity of the RO only "gradually emerge within a collaborative setting of inquiry" (1985:215).

In The Manufacture of Knowledge, a similar argument is ad- vanced by Knorr-Cetina (1981 : 3):

Rather than considering scientific products as somehow cap- turing what is, we will consider them as selectively carved out, transformed and constructed from whatever is. And rather than examine the external relations between science and the "nature" we are told it describes, we will look at those internal affairs o f scientific enterprise which we take to be constructive.

[On a realist account,] we tend to think of scientific "facts" as given entities, and not as fabrications. In the present study, the problem of facticity is relocated and seen as a problem of (laboratory) fabrication.

Regarding the supposed theory- and interpretation-neutral character of data points, and the blurred distinction between fact and artefact, the following discussion is relevant. In the late 1960s, Jocelyn Bell, a research assistant in the radio as t ronomy laboratory at Cambridge, observed "the persistent appearance of a strange section of ' scruff ' " on the recordings from equipment searching for quasars. The question was whether to interpret this scruff as fact or as instrument-generated artefact. For La- tour and Woolgar (1986:33) what is particularly relevant to a constructivist reading of science is

the method by which Bell made sense of a series of figures such that she could produce the account: "There was a recur- rence of a bit of scruff." The processes which inform the initial perception can be dealt with psychologically. However, our interest would be with the use of socially available procedures

126

for constructing an ordered account out of the apparent chaos of available perceptions. (Emphasis added.)

In addition, as the explanatory paradigms evolve so will descriptive accounts of what constitutes 'reality.' For example, RDs of the neural circuitry of the primate and human brain of the earth's weather systems, of the properties of pions, sister and daughter muons and neutrinos, etc., are contingent on existing theory, paradigmatic assumptions, technical recording apparatus (Latour and Woolgar's, 1986:51, 'inscription devices'), and so forth. The reality descriptions science generates are as much a function of these considerations as of Barnes' (1977:25) so-called "one world, one reality, 'out there,' ... the source of primitive causes."

Regarding such contingencies of theory, interpretation and apparatus on what we take to constitute reality and facticity, Knorr- Cetina (1981 : 33) seriously questions whether

the problem of facticity is to be located in the correspondence between the [cognitive] products of science and the external world ... The process of scientific enquiry ignored by objectiv- ism (its "context of discovery") is itself the system of reference which makes the objectification of reality possible ... Thus, the problem of facticity is as much a problem of the constitution of the world through the logic of scientific proce- dure as it is one of explanation and validation.

Consequently, Knorr-Cetina rejects any neat separation between the context of discovery and the context of explanation. Nor, as claimed earlier, is there a clear demarcation between, on the one hand, the reality-denoting element in RDs and, on the other, the conventional, constructivist and theory-contingent dimensions to scientific representations. Ironically, while clearly recognizing the constructivist component to RDs in science, Barnes would bifur- cate these RDs into: (i) their conventional/constructNist dimension as against (ii) their reality-denoting feature. Once we recognize (i) then reference to (ii) can constitute nothing more than a Kierkegaardian 'leap of (realist) faith.' It is not accidental that Barnes simply packs his claims re. (ii) into a few brief and unar- gued-for propositions. Once granting the importance of (i), why

127

does Barnes - or any other constructivist sociologist of science for that mat ter - feel it necessary to go on and make metaphysical claims re. the 'one reality out there, ' the supposed ultimate primitive cause of our empirical knowledge? Why not simply concen- trate on the conventional/constructivist dimension to RDs and scientific knowledge and leave the 'deeper' and, in my opinion, epistemologically unresolvable issues, to the philosophers?

There is the possible rejoinder to the stand taken here that any account of RDs in science which omits reference to what RDs ultimately represent would render totally unintelligible why inquirers find useful some RDs over others. Continuing the rejoinder, the reason why at least some RDs (e.g., the three-dimensional DNA model of Watson, Crick and Watkins or Pauli's model of the neutrino and its properties) function as such powerful predictive and explanatory tools in science is simply because such models more and more come to approximate reality through suc- cessive model revisions. This is a strictly pragmatic appeal to the heuristic value and utility of (at least some) RDs. The logic of the argument is impeccable i f one assumes the initial premise that a RD could have utility value if and only if it accurately or truth- fully (another favorite realist expression) represented what it portends to represent, namely, reality or at least some segment of reality.

But is this necessarily the case? The Ptolemaic, geocentric system of stellar, lunar and planetary representations is generally considered as inaccurate. Still, it was employed until the advent of the Loran system and satellites to constitute the basis for both maritime and aerial navigation - with surprising predictive accuracy and theoretical coherence. Dalton's and, later, Rutherford's model of the atom is another example. Both these macro- and micro-level models now belong to the history books. (For an interesting discussion of the cognitive maps and RDs employed by the ancient sea-fating Polynesians, see Oatley, 1977.)

An RD can therefore possess heuristic value without necessarily mapping in any truth-preserving fashion onto anything at all. That is, the RD need not preserve or capture anything about a so-called mind- and culture-independent reality. I do not have a ready reply to the realist's question, 'Why do some RDs have more heuristic value than others?' I am not questioning at this

128

point that some RDs do indeed have such value. However, I do not feel that the issue is resolved by such simplistic answers as, 'Well, it's their approximation to reality that grounds their explanatory and predictive utility!' The problem here is the term 'approximation.' Once we seriously question the pictorial metaphor re. the RD-RO relation, then have we said anything at all with such locutions as, 'approximation' (correspondence) to reality? Such locutions, and the language-game in which they are embedded, continue to draw on a simplistic picturing theory of representation.

4. Lessons to be learned

Such discussions and controversies continue to divert our attention from the interpretive factors in scientist's construction and reconstruction of RDs over time. Additionally, certain recent case studies in this area (e.g., Collins, 1981b, 1985; Garfinkel, Lynch and Livingston, 1981 ; Knorr, Krohn and Whitley, 1981 ; Knorr- Cetina, 1981, Knorr-Cetina and Mulkay, 1983; Latour and Wool- gar, 1986; and Lynch, 1985) show that inquirers' accounts of the supposed mind- and culture-independent reality of Barnes is certainly subject to conceptual revision over time. Obviously, for realists ontological assumptions constitute a sine qua non for understanding scientists' empirical and theoretical activities and apart from such a realist ontology such activities would make no sense at all. For constructivists, on the other hand, the issue is not the respective merits of one ontological framework over another but inquirers' discourse, accounts, cognitive constructions and praxis . Besides, as mentioned above, a serious problem with any realist ontology is that any set of reality-denoting claims to which inquirers are collectively prepared to attach evidential and existential import varies over time and is continually subject to negotiation and revision. Accord ing ly , it is s imp ly incons i s ten t w i th any thorough-going cons truc t iv i s t s trategy to ven ture into

realist - or perhaps even anti-realist - claims.

At least one philosopher of science has recently argued that what we take to constitute 'reality' changes with our representations and RDs. Accordingly, on this account no neat, simplistic

129

distinction is possible between Nature and (representational) Con- vention. Hacking ( 1 9 8 4 : 1 3 9 ) c o n c l u d e d from the suggestions of Kuhn and others that

with the growth of knowledge we may, from revolution to revolution, come to inhabit different worlds. New theories are new representations. They represent in different ways and so there are new kinds of reality. So much is simply a consequence of my account of reality as an attr ibute of representation.

Continuing, Hacking (1983:136) added that,

It will be protested that reality, or the world, was there before any representation or human language. Of course. But con- ceptualizing it as reality is secondary. First there is the human thing, the making of representations. Then there was the judg- ing of representations as real or unreal, true or false, faithful or unfaithful. Finally comes the world, not first but second, third or fourth.

One does not have to endorse all the claims of early-twentieth century positivism to see the merits o f avoiding the murky claims and even murkier arguments of the realist/anti-realist debate. Hacking (1983:145) similarly concludes that realism and anti- realism "scurry about, trying to latch on to something in the nature of representation that will vanquish the other. There is nothing there." I for one believe that a thorough-going constructivist account of representation and RDs in science is possible which avoids any entanglement with this unresolvable and debili- tating realist/anti-realist debate. Such an account would look quite different from that proposed, for example, by Barnes. (For a pragmatic statement of what such a disentangling strategy would look like, see Collins, 1981a, 1983a, for discussion of both the empirical and the radical programmes of relativism. For a defense of the latter programme, see Tibbetts, 1985, 1986, 1987.)

For one thing, it is clear that a constructivist/relativist account would not be commit ted - as is Barnes - to the pictorial account of representation discussed earlier. The pictorial metaphor for Barnes presumes some sort of orie-on-one mapping relation be-

130

tween RD and RO. With the emphasis now shifted to the RD and its contingent construction and revisions over time, and away from Barnes' ' independently-existing RO,' the language-game of mapping relations loses its logical force. For realists the pictorial model re. the R D - R O relation will still constitute a useful explanatory metaphor. However, keeping in mind my earlier criti- cisms (that the mapping model of Barnes is unable to express spatially- and temporally-dynamic properties), I would contend that even as a metaphor the language-game of mapping relations is seriously deficient. Additionally, there will be no talk about the extent to which RDs (whether two-, t h r ee -o r n-dimensional models, formulae, graphs, etc.) preserve or capture the so-called essential (structural) features of what is represented. What it is that is ultimately represented by RDs and with what degree of accuracy and fidelity will be of no concern for this proposed thorough-going constructivism.

I therefore conclude that there is no single nor fixed answer to the question, 'What after all do RDs in science represent?' Nor is the looked-for-answer located outside the domain of scientists' constructivist and interpretive activities concerning both what is being represented and h o w it is best represented. The realist's purported distinction between Nature and Convention can only be sustained by ignoring the hermeneutics of scientific representation.

References

Barnes, B. (1977). Interests and the growth of knowledge. Boston: Rout- ledge and Kegan Paul.

Barnes, B., and Edge, D., Eds. (1982). Science in context: Readings in the sociology of science. Cambridge: MIT Press.

Bhaskar, R. (1978). A realist theory of science. New York: Humanities Press. Bhaskar, R. (1981). Realism. In W. Bynum et al. (Eds.), Dictionary o f the

history of science, 362-363. Princeton: Princeton University Press. Bloor, D. (1976). Knowledge and social imagery. Boston: Routledge and

Kegan Paul. Bohme, G. (1977). Cognitive norms, knowledge interests and the constitu-

tion of the scientific object. In E. Mendelsohn, P. Weingart and R. Whit- ley (Eds.), The social production of scientific knowledge, 129-141. Dordrecht: Reidel.

131

Collins, H. (1981a). What is TRASP?. The radical programme as a method- ological imperative. Philosophy of the Social Sciences 11 (2): 215- 224.

Collins, H., Ed. (1981b). Knowledge and controversy: Studies of modern natural science. Special issue of Social Studies o f Science 11 : 1-158.

Collins, H. (1983a). An empirical relativist programme in the sociology of scientific knowledge. In K. Knorr-Cetina and M. Mulkay (Eds.), Science observed: Perspectives on the social study of science, 85-114. London: Sage.

Collins, H. (1983b). The sociology of scientific knowledge: Studies of con- temporary science. Annual Review of Sociology 9: 265-285.

Collins, H. (1985). Changing order. London: Sage. Collins, H., and Pinch, T. (1982). Frames of meaning. Boston: Routledge

and Kegan Paul. Garfinkel, H., Lynch, M., and Livingston, E. (1981). The work of a dis-

covering science construed with materials from the optically-discovered pulsar. Philosophy of the Social Sciences 11(2): 131-158.

Grenier, M. (1983). Cognition and social construction in laboratory science. Society for Social Studies of Science Review 1(3):2-16.

Hacking, I. (1983). Representing and intervening. New York: Cambridge University Press.

Hollis, M. (1982). The social destruction of reality. In M. Hollis and S. Lukes (Eds.), Rationality and relativism, 67-86. Cambridge: MIT Press.

Hollis, M., and Lukes, S., Eds. (1982). Rationality and relativism. Cam- bridge: MIT Press.

Jarvie, I. (1983). Rationality and relativism. British Journal o f Sociology 34(1):44-60.

Jarvie, I. (1984). Rationality and relativism. Boston: Routledge and Kegan Paul.

Knorr, K., Krohn, R., and Whitley, R., Eds. (1981). The social process of scientific investigation. Dordrecht: Reidel.

Knorr-Cetina, K. (1981). The manufacture o f knowledge: An essay on the constructivist and contextual model of science. New York: Pergamon.

Knorr-Cetina, K. (1983). The ethnographic study of scientific work: To- wards a constructivist interpretation of science. In K. Knorr-Cetina and M. Mulkay (Eds.), Science observed: Perspectives on the social study of science, 115-140. London: Sage.

Knorr-Cetina, K., and Mulkay, M., Eds. (1983). Science observed: Perspec- tives on the social study of science. London: Sage.

Latour, B., and Woolgar, S. (1986). Laboratory life, 2nd ed. Princeton: Princeton University Press.

Laudan, L. (1977). Progress and its problems. Berkeley: University of Cali- fornia Press.

Laudan, L. (1981). The pseudo-science of science? Philosophy of the Social Sciences 11(2): 173-198.

Lynch, M. (1985). Art and artifact in laboratory science. Boston: Routledge and Kegan Paul.

132

Mendelsohn, E., Weingart, P., and Whitely, R., Eds. (1977). The social production of scientific knowledge. Dordrecht: Reidel.

Oatley, K. (1977). Inference, navigation and cognitive maps. In P.N. John- son-Laird and P.C. Wason (Eds.), Thinking: Readings in cognitive science, 537-547. New York: Cambridge University Press.

Peirce, C. [ t 878 ] (1931-35). Collected papers. Cambridge: Harvard Univer- sity Press.

Picketing, A. (1984). Constructing quarks. Chicago: University of Chicago Press.

Popper, K. (1972). Objective knowledge. New York: Oxford University Press.

Tibbetts, P. (1985). In defense of relativism and the strong programme. British Journal of Sociology 36(3): 471-476.

Tibbetts, P. (1986). The sociology of scientific knowledge: The constructivist thesis and relativism. Philosophy of the Social Sciences 16(1): 39-57.

Tibbetts, P. (1987). The sociology of scientific knowledge: Problems of knowledge-generation and -legitimation. Unpublished manuscript.

Woolgar, S. (1983). Irony in the social study of science. In K. Knorr-Cetina and M. Mulkay (Eds.), Science observed: Perspectives on the social study of science, 239-266. London: Sage.

Documents

Representation and the realist-constructivist controversy